adding PyData Global 2025 presentation link
Browse files
README.md
CHANGED
|
@@ -9,16 +9,19 @@ This repository provides a detailed guide and implementation of the Transformer
|
|
| 9 |
For implementions of more recent architectural innovations from DeepSeek, see the **Related Implementations** section.
|
| 10 |
|
| 11 |
## Table of Contents
|
| 12 |
-
1. [Summary and Key Insights](#summary-and-key-insights)
|
| 13 |
-
2. [Implementation Details](#implementation-details)
|
| 14 |
-
- [Embedding and Positional Encoding](#embedding-and-positional-encoding)
|
| 15 |
-
- [Transformer Attention](#transformer-attention)
|
| 16 |
-
- [Feed-Forward Network](#feed-forward-network)
|
| 17 |
-
- [Transformer Decoder](#transformer-decoder)
|
| 18 |
-
- [Encoder-Decoder Stack](#encoder-decoder-stack)
|
| 19 |
-
- [Full Transformer](#full-transformer)
|
| 20 |
-
|
| 21 |
-
|
|
|
|
|
|
|
|
|
|
| 22 |
|
| 23 |
## Quick Start
|
| 24 |
View the complete implementation and tutorial in the [Jupyter notebook](Transformer_Implementation_Tutorial.ipynb).
|
|
@@ -216,6 +219,24 @@ These visualizations help understand the inner workings of the transformer and v
|
|
| 216 |
|
| 217 |
For detailed code and interactive examples, please refer to the complete implementation notebook.
|
| 218 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 219 |
## Related Implementations
|
| 220 |
|
| 221 |
This repository is part of a series implementing the key architectural innovations from the DeepSeek paper:
|
|
|
|
| 9 |
For implementions of more recent architectural innovations from DeepSeek, see the **Related Implementations** section.
|
| 10 |
|
| 11 |
## Table of Contents
|
| 12 |
+
1. [Summary and Key Insights](## summary-and-key-insights)
|
| 13 |
+
2. [Implementation Details](# implementation-details)
|
| 14 |
+
- [Embedding and Positional Encoding](### embedding-and-positional-encoding)
|
| 15 |
+
- [Transformer Attention](### transformer-attention)
|
| 16 |
+
- [Feed-Forward Network](### feed-forward-network)
|
| 17 |
+
- [Transformer Decoder](### transformer-decoder)
|
| 18 |
+
- [Encoder-Decoder Stack](### encoder-decoder-stack)
|
| 19 |
+
- [Full Transformer](### full-transformer)
|
| 20 |
+
- [Testing](### testing)
|
| 21 |
+
- [Visualizations](### visualizations)
|
| 22 |
+
3. [PyData Global 2025 Presentation](## PyData Global 2025 Presentation)
|
| 23 |
+
4. [Related Implementations](##Related Implementations)
|
| 24 |
+
|
| 25 |
|
| 26 |
## Quick Start
|
| 27 |
View the complete implementation and tutorial in the [Jupyter notebook](Transformer_Implementation_Tutorial.ipynb).
|
|
|
|
| 219 |
|
| 220 |
For detailed code and interactive examples, please refer to the complete implementation notebook.
|
| 221 |
|
| 222 |
+
## PyData Global 2025 Presentation
|
| 223 |
+
For those of you who prefer to learn from videos, you can watch my PyData Global 2025 presentation "[I Built a Transformer from Scratch So You Don’t Have To](https://www.youtube.com/watch?v=ID5zSzycQBg)"
|
| 224 |
+
|
| 225 |
+
• How the original Transformer architecture works
|
| 226 |
+
|
| 227 |
+
• How to translate each component into PyTorch
|
| 228 |
+
|
| 229 |
+
• Key ideas: attention, masking, positional encoding, FFN
|
| 230 |
+
|
| 231 |
+
• A decoder-only forward pass, step-by-step
|
| 232 |
+
|
| 233 |
+
• Common implementation bugs — and how to debug them
|
| 234 |
+
|
| 235 |
+
• Where to go next (code, tutorials, training references)
|
| 236 |
+
|
| 237 |
+
|
| 238 |
+
|
| 239 |
+
|
| 240 |
## Related Implementations
|
| 241 |
|
| 242 |
This repository is part of a series implementing the key architectural innovations from the DeepSeek paper:
|